Традиционно российская школа мысли характеризовалась тенденцией к универсальным подходам и обобщениям, стремясь охватить связь таких понятий и явлений как Мышление, Вселенная и Логос. Она породила множество строгих математических теорий а также красивых моделей и загадочных гипотез.
И.П.Павлов, Л.С.Выгодский, А.Н.Леонтьев, Д.А.Поспелов представляют собой те ясные звезды на российском научном горизонте, глубокая интуиция, потенциал и творческий вклад которых существенно стимулируют глобальное видение механизмов человеческого мышления и гармоничное развитие исследований по естественному и исскуственному интеллекту. Они подготовили плодотворную почву для российской школы искусственного интеллекта, позволившую ей занять почётное место среди зарубежных школ искусственного интеллекта.
открытые под руководством Д.А.Поспелова, привели к разработке принципов Когнитивной Относительности, Рациональности и Ясности (CRRC)
. Они обеспечили создание само-организующихся Систем Взаимодействующих Контекстных Систем (C2S)
способных моделировать поведение сложных интеллектуальных систем и явлений. Это привело к созданию новой концепции Эволюционно Вовлекающих Интеллектуальных Систем (EI)
, основанных на λ-интеллекте
From Universal Scales to Systems of Communicating Contextual Systems
IIAT, 11, rue des Sablons, Brussels, 1000, Belgium
1. How does mind perceive and anticipate reality?
Being impressed by rigorous abilities of fundamental sciences to reconstruct reality and generate useful predictions, the human mind does not very often question itself why and how these sciences were constructed and how the change in perspective of mind observations might influence the rigor of those sciences.
According to Western views, in order to create a picture of reality the mind builds concepts and classifications. It tends to distinguish, classify, compare, measure and verbalize. To discover particular laws of behavior, theories are constructed by varying some parameters while keeping others constant. Also models take into account only a few variables while omitting others.
According to Eastern views knowledge lies beyond concepts, classifications and words, and humans can only reach reality through their direct experience. From this point the world exists as a whole of opposites, which are not against each other, but which compliment each other: white creates black; light creates dark; mountains create valleys. Indeed while conceptualizing it also cuts reality into smaller pieces, but a concept and its opposite are parts of a unified whole and the motivation of mind is to discover and clarify their unity.
Traditionally Russian schools of thought were deeply concerned with both views. They were strongly characterized by a tendency to create and generalize a universal vision, which would grasp not only a notion of Mind, but also a notion of Logos and even a notion of the whole Universe. They produced diversity of the rigorous theories as well as of the mysteriously beautiful models and hypothesis. Pavlov, Leontiev, Vugodsky, Pospelov and many others represent those lucid stars on the Russian scientific horizon, whose deep intuition, courage and creativity essentially stimulated a harmonized global vision on human mind mechanisms and functionality of natural intelligence. This resulted in a blossoming tree of the fruitful Russian schools of AI.
2. Historic and Interdisciplinary view
From the very beginning of Artificial Intelligence, the problem of the AI subject and its relation to natural intelligence and mind was the most often attracted and discussed. While most sub-disciplines of computer science seem to be defined by their methods, AI seems more to have been defined by the source of its inspiration. In 1950, Alan Turing published his famous paper "Computing Machinery and the Mind", which can be metaphorically described as posing the problem of how a computer could distinguish a male from a female merely by asking questions. This problem has become known as the "Turing Test" and has served as an inspiring vision and philosophical charter for many years since the beginning of AI.
The pioneers of the field, such as Simon, McCarthy, Minsky and Feigenbaum, were in fact guided by the study of human intelligence, and sought deeper understanding of human cognition with the hope that this effort would lead to better machines. According to McCarthy "... human level artificial intelligence requires equipping a computer program with a philosophy. The program must have built into it a concept of what knowledge is and how it is obtained." Finally he pointed out that he considers his main purpose to be the development of a theory of "contexts". This term arises, as a necessary consequence when one wants to deal with knowledge properly. McCarthy thus also believes that "Mind has to be understood one feature at a time".
As a practical matter, Sloman offers similar views, for example, that "...'mind' is a cluster concept referring to an ill defined collection of features, rather than a single property that is either present or absent." Both Sloman and McCarthy also talked of "stances" which are levels of analysis one may make of a system, such as physical, intentional, design and function. Both assert that philosophy and AI still have much to offer each other.
However Hayes defined AI as "the engineering of cognition based on computational vision, which runs through and informs all of cognitive science". Hayes characterized the vision of "making artificial super-humans" as the initial goal of the pioneers of the field, Feigenbaum, McCarthy and Minsky. He considers the Turing Test as harmful and argues that the Turing Test now leads the field to disown and reject its own success. Hayes argued that AI should not be defined as an imitation of human abilities. He believes that if we abandon the Turing Test vision, then "... the goal naturally shifts from making artificial super-humans which can replace us to making super-humanly intelligent artifacts which we can use to amplify and support our own cognitive abilities..." At the same time, the mainstream of researchers in AI simply understand the task of AI as engineering of useful methods (artifacts), without even much reflection on whether these take account of human attributes, such as mind, natural intelligence or cognition.
Despite the mainstream engineering, the cognitive orientation of AI remains a force in the field. From the very beginning the Russian school of AI was very much in concern with human attitudes of AI. It succeeded to combine western and eastern views and had produced its own way of scientific approach. Russian culture introduced diversity of intermediate notions and patterns (such for example as the notion of Pravda complementing the notions of Truth and False), as well as numerous outstanding mathematical statements and algorithms (such for example as Lobachevsky geometry), which essentially stimulated development of both views.
It was more then four decades ago when Dmitry Pospelov began his inspiring study of the Semiotic Systems, Situated Logics, Universal Scales and Spaces.
His interest in psychology and neurology, in mathematical logics and fuzzy sets, in linguistics and behavior sciences had been stimulated the blossoming tree of a broad Russian school of AI. His typical way of approaching constructive model was formalized as a cortege, or train (or even a simple list) of elements, each of which then may be represented well in a traditional way. This reflected his original flexibility, profound vision and interdisciplinary views.
His intuition was deeply based on a belief that Osgood scales and related spaces may lead to a better understanding of semantically grounded systems. This finally has lead to a discovery and development of the Universal Scales
(33, 34, 5, 6). Latter research in this direction allowed development of the unified integrating framework for modeling a diversity of cognitive and complex real phenomenon. The basic principles of Cognitive Relativity, Rationality and Clarity
(10, 13) were crystallized to underline this direction of the Russian school of thought. It became clear that both views can be integrated on the basis of these principles. The mathematical theory of Systems of Communicating Contextual Systems
is based on recursive mechanisms of theorem proving and constraints recognition and satisfaction, the first elements of which were also developed in 1974-1978 under the supervision of Dmitry Pospelov, and in a productive collaboration with other Russian mathematical schools such as of Prof. Maslov (24) and of Prof. Kotov (23).
The Contextual theory of Cognitive States
(5-13) and the Systems of Communicating Contextual Systems (C2S)
suggest a unified framework for modeling life-cycles of patterns, representations, and of possible ways of their construction, generation, interaction and transformation (14). This framework allows modeling complex center-activated or distributed self-organizing phenomenon, which may have centered or distributed cognition. It allows invention of a new kind of AI systems, Evolutionary Evolving Intelligent Systems (EI)
, which are based on what we call by λ-Intelligence
, and which are principally open and flexible, continuously learning, self organized, cognitively tailored and collectively adaptive systems (10-15).
The three basic principles, - cognitive relativity, rationality
underline any-time, any-place and any-role modeling of Evolutionary Evolving
(10). Their understanding helps to resolve many of the puzzles of human
activities, including: intellectual, intuitive, creative, emotional, psychological, social and
cultural. They help to explain the origin, role and evolution of symbolic representations,
languages, cultures and histories. They provide a constructive mechanism for every day smart
intelligent systems management, communicating support activities and strategic planning in dynamic
environments. The rest of the paper will shortly discuss these core principles and how they can be
used to build Contextual Spaces, Contextual Systems
and self-organizing Systems of Communicating
. Thus the notion of the Universal Scales
is the underlying idea for a design of
Evolutionary Evolving Intelligent Systems
3. From Physiological Studies to Cultural Sciences
At the end of his life, I. Pavlov discovered the existence of "dynamic stereotypes" (DS) in brain activity (28). By doing so he changed the paradigm of the simple theory of conditioned reflex, to the idea that instead of reacting only to outside stimulus, the brain reacts to internally reflected DS (patterns) of previously experienced stimulus. This also allowed him to draw a hypothesis on the emotional relationship of subjects to their external reality. His work first suggested some physiological evidence of an important role of subjective representations and transformations by the brain of reality. Anohin followed this idea, suggesting a Theory of Functional Systems (2), which underlines an important role of informational processes together with physico-chemical processes in brain activity.
It has become clear that subjective mappings (imprints, patterns) of reality "are formed under the dominant motivation, which may be best understood as a holographic canvas, on which outside stimulus from different parameters of reality build a mosaic of informational dynamic stereotypes" (41, page 6). Such dynamic stereotypes, or dynamic internally reflected patterns of reality, bring light on understanding different sides of brain and mind activities, ranging from unconscious brain activities to conscious mind involvement.
New results in neurosciences, bio-intelligence as well as genetic and systemic biology allow advances in further understanding and modeling the interacting roles of brain, mind, memory, body, emotion, behavior and robust bio-rationalities. Recent results from cognitive and cultural sciences consider cognition as dependent on recognized patterns, which have been reflected and figured out from experiences, previous interactions and situations. They also suggest that what individuals have in their minds are different kinds of patterns representations, and that mutual representations are possible due to experiencing similar situations, relations, roles, payoffs, etc.
Such clarified representations influence memories, attention, as well as dominant motivations. Their structure and granulation intends to be modified to become the most appropriate for possible implications and applications. Structures of mutual representations may evolve organizations of actors, which then may be used to control individual actors, their modes of perception, cognition, behavior and comprehension. These mutual representations allow formalization of different forms of collective, social and cultural activities.
4. Universal Scales and λ-Intelligence
Information technology evolved in the last several decades in part, to model human intelligence and cognition in the machine. Much of physical theory became comprehensible in a single framework, once it was recognized that Hilbert Spaces provided a basis for representation of mathematical objects useful to that theory (19). Operators describing relationships of physical theory are thus now often represented as operators in and on Hilbert Spaces.
In the early 70s Lotfi Zadeh suggested the notion of fuzzy sets and the concept of linguistic variables (43-45), which had stimulated the search of constructive answers to many important questions. One question among those was: are there any universal tools, hidden in human minds or languages, which act as universally functional cognitive figures, which would allow distinguishing of useful patterns of the reality. The latter research showed that pointed by Zadeh linguistic terms such as small, big, etc. may be observed as such cognitive figures, when they are represented on the Universal Scales for measuring degrees of distinguishability. While a "small men" or a "small planet" have different meanings on traditional physical scales (kg, m, etc), they may have the same representation on the Universal Scales.
Functions of Experience
were defined to map real values to their cognitive meanings on the Universal scales.
It allowed meanings of terms such as small, big, etc as cognitive functional quarks
, which may bring light on the origin and meaning of other symbolic figures from different mutual representations and languages. Consider for example a definition of "armchair" as the same as the "sofa" but short in the length.
The number of distinguishing figures is channeled by limits of human physiology and it is the natural consequence of a specific human experience. When multiplying by a huge number of possible contexts it may result in a potentially unlimited number of real meanings. It became clear that the universal scales suggest the most relevant scales for a diversity of computations and that many of those computations, which are traditionally carried on physical scales, have to be settled on the Universal Scales to avoid wrong interpretation, where difference in contextual interpretations may be neglected.
This work was partly published in 1975-2005 (Pospelov, Ezhkova, Bianco and others). Figures 1-5 illustrate some of those operations and some basic features of Functions of Experience. More detailed discussion can be found in (33-34, 5-6). Technically, the Theory of Distinguishability constructs a vector space of observers, on which operators relevant to AI forms of analysis can be constructed, which observables and operators follow definable rules of logic on the spaces, and which develop and apply specified measures on the observations. This framework allows one to relate available experiences to appropriate spaces of measurements, symbolic representations, knowledge and other patterns in systematic and productive ways.
provide a simple and reliable tool for measuring distinguishability. However to be able to distinguish we need to know a frame of reference, where things become really clear and different. This frame may be a game, an activity, a culture, a situation, a view, or in general a context
, within which things, roles, shapes, lines, trajectories, signs, and patterns may be distinguished.
Consider the definition "An armchair is the same as a sofa, but shorter". Things are often defined first by similarity with known things, and then by distinguishability from others. However it is not enough just to compare two things, or their values, or even their vicinities, but it is necessary to take into account the whole context, within which this comparison may occur. The same two objects, already distinguished, may be observed as very distant or as very close depending on contexts. Classification also results from ability to distinguish. Useful classifications and ontology have to be also functional, dynamic and context-sensitive.
The logical mechanism of Systems of Communicating Contextual Systems
is based on λ-intelligence
, which is a recursive mechanism of theorem proving with constraints acquisition and satisfaction. λ-intelligence
recognizes a functional form of negation in a following way "Everything is possible by exception what is not yet possible"
(10-15). It constitutes a very useful mathematical platform to handle with a notion of Contextual and multi-contextual negation, and provides an effective tool for self-organization of Communicating Contextual Systems, of their static as well as dynamic abilities. The roots of λ-Intelligence
can be found in the earlier works on recursive theorem proving and planning, which were stimulated and supervised by Dmitry Pospelov in 1969-1975 (5).
Systems of Communicating Contextual Systems (C2S)
represent the constructive self-tuning continuously learning mechanism of acquisition, representation and processing of information and knowledge in complex environments. They allow construction of Evolutionary Evolving Intelligence
. They may represent and utilize related experience based on three basic principles of Cognitive Relativity, Rationality and Clarity
Evolutionary Involving Intelligence Systems
, based on Cognitive Relativity, Rationality and Clarity
are recognized now as the Russian school of Artificial Intelligence. λ-intelligence
is evolved to predict possible organizational forms of embodiment of C2S.
This research is essentially based on the Universal Scales
, developed under the supervision of Dmitry Pospelov.
Fig. 1. Universal Scales: Functions of Experiences
Fig.2. Symbolization: Interpretation, Verbalization, Restoration, Translation
Fig.3. Gradients and Differentiation